NIC Processor using variable precision block quantization

Abstract

A nearly instantaneous companding processor uses variable precision block quantization for reducing the digital transmission rate of speech signals while at the same time maintaining a high quality signal in a digital communication system. The variable number of bits which are used to encode each sample is controlled as a function of the maximum sample magnitude in a block of N samples. This technique takes advantage of the amplitude probability density function of a speech signal to thereby reduce the quantizing noise without increasing the transmission rate.

What is claimed is: 
     
       1. An apparatus for encoding a block of N digital samples having sign and magnitude bits, being of the type wherein each of said N samples are quantized in a quantizer and the quantizer output is then encoded for transmission, said apparatus comprising: maximum value finding means for finding the maximum value of said N samples;   bit number determining means for determining, as a function of said maximum value, the number of bits to be used in encoding said N samples;   step size determining means for determining, as a function of said maximum value, a quantizing step size for said N samples;   quantizing means for quantizing said samples in accordance with said step size to obtain quantized values; and   means for encoding said quantized values to produce N encoded samples each encoded with a number of bits determined by said bit number determining means.   
     
     
       2. The apparatus of claim 1 wherein each of said N samples are represented by a plurality of bits in a data compressed format having a predetermined number of segments, and wherein said bit number determining means examines a group of bits from said maximum value, the number of bits in said group being the minimum number of bits which can be used to uniquely identify each of said segments. 
     
     
       3. The apparatus of claim 2 wherein said bit number determining means comprises a memory device receiving said group of bits from said maximum value for determining said bit number in accordance with said group of bits. 
     
     
       4. The apparatus of claim 3 wherein said group of bits consists of three bits, and said bit number is one of 2, 4 and 6 bits, excluding said sign bits. 
     
     
       5. The apparatus of claim 4 wherein said group of bits are taken from the most significant magnitude bits in said maximum value, and said memory device provides the input/output function:   ______________________________________                                    
Group of Bits                                                             
            Bit Number Excluding Sign Bit                                 
______________________________________                                    
0 0 0       2                                                             
0 0 1       2                                                             
0 1 0       2                                                             
0 1 1       2                                                             
1 0 0       4                                                             
1 0 1       4                                                             
1 1 0       6                                                             
1 1 1       6.                                                            
______________________________________                                    
     
     
     
       6. The apparatus of any one of claims 1-5 wherein said step size determining means determines said quantizing step size as a function of said bit number as determined by said bit number determining means. 
     
     
       7. The apparatus of claim 6 wherein said step size determining means comprises a divider for dividing said maximum value by a power of 2 to produce said quantizing step size. 
     
     
       8. The apparatus of claim 7 wherein said power of 2 is equal to   2.sup.n i.sup.-1,     where n i  -1 represents said bit number excluding said sign bits.   
     
     
       9. An apparatus for decoding a plurality of blocks of data, each block having N digital samples, each of said samples having a number of bits which is variable from block to block but is the same for all samples in any given block, each of said blocks of data including protocol information indicative of the value of the largest of said N samples, said apparatus comprising: protocol extracting means for extracting said protocol information from said blocks of data;   sample forming means for forming said N digital samples in accordance with said value of the largest of said N samples represented by said protocol information; and   decoder means for decoding said N digital samples in accordance with said protocol information to provide decoded values.   
     
     
       10. The apparatus of claim 9 comprising sample length determining means responsive to said extracted protocol information to provide sample length information to said sample forming means indicating the number of bits in each sample in accordance with said value of said largest of said N samples. 
     
     
       11. The apparatus of claim 10 wherein said decoder decodes each of said N digital samples as a function of (i) the value of each of said N samples, (ii) said value of the largest of said N samples, and (iii) the value of said number of bits in each sample. 
     
     
       12. The apparatus of claim 11 wherein said decoder decodes said N digital samples in accordance with the function:   S.sub.j =[2(M-1)+1]S.sub.max /2.sup.n i,     where S j  =the decoded value, M-1=the value of the associated digital sample produced by said sample forming means, S max  =the value of said largest of said N samples, and n i=  the length in bits of said associated digital sample.   
     
     
       13. A process for encoding a block of N digital samples having sign and magnitude bits, the process comprising: finding the maximum value of said N samples;   determining the number of bits to be used in encoding said N samples as a function of said maximum value;   determining a quantizing step size for said N samples as a function of said maximum value;   quantizing said samples according to said determined quantizing step size to obtain quantized values; and   encoding said quantized values to produce N encoded samples each having said determined number of bits.   
     
     
       14. The method of claim 13 wherein said N digital samples are each represented by a plurality of bits in a compressed format having a predetermined number of segments, said step of determining said number of bits comprising examining a group of bits from said maximum value, the number of bits being examined corresponding to the minimum number of bits which can be used to uniquely identify each of said segments. 
     
     
       15. The method of claim 14 wherein said number of examined bits is three bits and said step of determining said number of bits further comprises selecting the number of said bits to be used in encoding said samples to be equal to one of 2, 4 and 6 bits, excluding sign bits. 
     
     
       16. The method of claim 15 wherein said selecting step comprises selecting the bit number excluding said sign bit as a function of the value of said group of bits as follows:   ______________________________________                                    
Group of Bits                                                             
            Bit Number Excluding Sign Bit                                 
______________________________________                                    
0 0 0       2                                                             
0 0 1       2                                                             
0 1 0       2                                                             
0 1 1       2                                                             
1 0 0       4                                                             
1 0 1       4                                                             
1 1 0       6                                                             
1 1 1       6                                                             
______________________________________                                    
     
     
     
       17. The method of any one of claims 14, 15 or 16, wherein said step of determining said quantizing step size comprises determining said quantizing step size as a function of said determined bit number. 
     
     
       18. The method of claim 17, wherein said step of determining said quantizing step size comprises dividing said maximum value by a power of 2 to produce said quantizing step size. 
     
     
       19. The method of claim 18 wherein said power of 2 is equal to   2.sup.n i.sup.-1,     where n i  -1 represents said determined bit number excluding sign bits.   
     
     
       20. A method for decoding a plurality of blocks of data, each block having N digital samples, each of said samples having a number of bits which is variable from block to block but is the same for all samples in any given block, each of said blocks of data including protocol information indicative of the value of the largest of said N samples, said method comprising: extracting said protocol information from said blocks of data;   forming said N digital samples in accordance with said value of said largest of said N samples indicated by said protocol information; and   decoding said N digital samples in accordance with said protocol information to provide decoded values.   
     
     
       21. The method of claim 20 wherein said forming step includes the step of determining the number of bits in said samples in response to said extracted protocol information, and forming said N digital samples in accordance with said determined number of bits. 
     
     
       22. The method of claim 21 wherein said decoding step comprises decoding each of said N digital samples as a function of (i) the value of said each sample, (ii) the value of the largest of said N samples, and (iii) the value of said determined number of bits. 
     
     
       23. The method of claim 22 wherein said decoding step comprises decoding each of said N digital samples in accordance with the function   S.sub.j =[2(M-1)+1]S.sub.max /2.sup.n i,     where S j  =the decoded value, M-1=the value of said each digital sample, S max  =the value of said largest of said N samples, and n i  =said determined number of bits.